Morpheus parser code — a morphological analyzer for Greek and Latin.

Build everything (cruncher, tools, and stem libraries) from the project root:

make

This compiles the C source in src/, installs binaries to bin/, and builds the Greek and Latin stem libraries in stemlib/.

To rebuild just the C source without the stem libraries:

make build

make test

This runs the full test suite: cruncher output tests for all supported flag combinations, followed by a from-scratch rebuild of both stem libraries to verify the build pipeline. Test baselines are not tracked in git. To generate or update them:

make baselines

Baselines capture the current cruncher output so that future changes show up as diffs. A diff is reported but does not count as a failure — only crashes and build errors do.

MORPHLIB=stemlib bin/cruncher < wordlist > crunched

When reading from stdin, analyses are written to stdout and failed words and statistics are written to stderr. Alternatively, an input filename can be given as an argument (a .words extension is appended), in which case output goes to .morph, .failed, and .stats files.

By default, the cruncher outputs in Perseus format.

-P disables Perseus format, falling through to a human-readable display that shows each word with its possible lemmas and full morphological breakdowns using $..& beta-code markup.

-d selects database format: labeled lines (:raw, :workw, :lem, :prvb, :aug1, :stem, :suff, :end) with the morphological flags for each morpheme. Implies -P.

-x selects lexicon format: one tab-delimited line per analysis with rawword, workword (if different), lemma, morphological flags, and crasis. Must be combined with -P to take effect (Perseus format takes priority otherwise).

-e changes the morphological flags to a numeric form-info value. In Perseus format (the default), this replaces the text flags in <NL> lines. Combined with -P, it selects its own output format showing stem type, ending string, paradigm info, and word form.

-p selects the :summ/:stem/:end format (like -d but more compact, with a one-line :summ summary per analysis). Also collapses whitespace in the detail lines. Must be combined with -P to take effect.

-l outputs lemma information only: form: followed by the input word, then one line per unique lemma. Overrides all other format flags.

-L sets the language to Latin (default is Greek).

-I sets the language to Italian.

-S disables strict case matching (enabled by default). Allows analysis of capitalized words by lowercasing them, useful for sentence-initial words.

-n enables accent-insensitive analysis. If normal analysis fails, retries with smooth and then rough breathing variants.

-V restricts analysis to verb forms only, skipping nominal and indeclinable checks.

-k keeps beta-code transliteration in the output. Only has effect with -P (in the human-readable format, controls whether output is converted to smart-alpha).

-T disables performance timing (timing is on by default). Suppresses the :longtime and :time progress messages on stderr.

-o FILE writes output to FILE (with .failed and .stats companion files). Use -o - to explicitly route output to stdout and failures/stats to stderr.